John Rawls' Primary Goods Approach to Justice: An Analysis of Indexing Problems

There has been extensive debate over the past few decades regarding the criteria by which we should measure distributive justice. In conceiving a just state of affairs it is imperative that we determine the most appropriate measure of the distributions we are evaluating. Different approaches have their own merits, and range from evaluations of the distribution of happiness, to that of wealth, to that of life prospects. John Rawls and fellow Rawlsians believe that primary goods are most representative of citizens’ fundamental interests. According to Rawls (1999: 54), primary goods are those that individuals prefer to have more of rather than less, and include “rights, liberties, opportunities, income and wealth.” To this end, Rawls’ conception of justice as put forward in his 1971 book ‘A Theory of Justice’ (TJ) uses primary goods as the fundamental metric of justice.

This essay aims to critique Rawls’ primary goods approach, beginning with a short description of the main alternative theory, namely Sen’s capabilities approach. It will then focus on the indexing problems that occur when trying to implement it. The first, indexing problem 1, refers to the issue of how to aggregate a bundle of primary goods, whilst the second, indexing problem 2, considers the competing methods of ranking the social states of primary goods distributions. Finally, the essay will discuss the difficulty involved in constructing such an index, with an explanation of how Rawls fails to adequately tackle it.

The Capabilities Approach

Sen (1979) presented his account of capabilities as an alternative to the Rawlsian approach. Sen believes that Rawls failed to take sufficient note of the diversity of individuals and therefore their relative abilities to convert primary goods into their own account of the good. For Sen (1979: 215-216), Rawls’ approach “is not merely ignoring a few hard cases, but overlooking very widespread and real differences” as individuals vary in their “health, longevity, climatic conditions, location, work condition and even body size.” Sen takes the case of a severely disabled individual as an example. If we focus only on holdings of primary goods, this individual would not be granted any redistribution on the grounds of his impairment according to Rawls’ difference principle. The crux of Sen’s critique goes further than the case of disabled people, but more broadly expresses concerns that primary goods holdings alone are an inadequate metric of justice given people’s interindividual differences. This, Sen (1990: 112) argues, can result in “unjustified inequality and unfairness.” What we should be focussing on is people’s capabilities to function. For Sen, the real opportunities that an individual has to pursue her own life plan depend not only on her holdings of primary goods, but also on a variety of factors that influence the extent to which she can use her primary goods to useful effect.

The primary purpose of this essay is not to evaluate the debate between primary goods theorists and capabilities proponents. However, the fact that over 40 years has passed since TJ was published, with no resolution to the debate, shows that there may be no definitive answer to the ideal metric of justice. The remainder of this essay will explain, how, even if we believe primary goods to be the best metric of justice, there is a stubborn indexing issue with primary goods that appears to have no obvious solution.

Rationality and ‘Indexing Problem 1’

Rawls (1999: 10) states that his principles of justice are those that “free and rational persons concerned to further their own interests would accept.” Behind the hypothetical veil of ignorance, agents are ignorant of their: place in society, class position, social status, fortune in the distribution of natural assets, and their conceptions of the good (p.11). In this state they are said to be in the original position (OP). Thus the OP works to ensure agents are in a state of equality, as ignorance of such information renders it impossible for an agent to pursue her own ends. As a result, derived principles of justice will be uninfluenced by personal considerations that are morally irrelevant in the formation of social principles.

In the society that agents are forming the structure of, there are n individuals, where x_i is individual i’s bundle of primary goods, and i = 1,…,n. The job of agents in the OP is to use these bundles to decide upon a just social state of primary goods, X = (x₁,x₂,…,x_n).

As there will be several possible social states to decide upon, agents need to be able to rank the options for particular social states based on their preferences. In other words, they must be ‘rational’. Returning to the quote at the beginning of this section, it is a confident move by Rawls to associate rationality with agents who, behind the veil, have a serious lack of information as to the society in which they will reside. The theory of rational choice necessitates that individuals have sufficient information from which they can rank certain outcomes in accordance with their preferences (Osborne, 2004: 5). The problem for Rawls is that if agents behind the veil are made ignorant to the extent that they are unaware of their own conceptions of the good, then how may they form preferences? This is a flaw in Rawls’ work and one that he caters for by adding the assumption that agents do possess a “thin theory of the good” (Rawls, 1999: 349). In particular this involves a common preference to hold more rather than less primary goods.

Now that I have explained that agents must have preferences in order to choose a social state, we must now ask what exact preferences agents will have. Rawls contends that agents, given their ignorance of their position in society, will opt for a maximin strategy whereby social states are ranked according to the welfare of the least well-off individual. This is the strategy that underpins the difference principle. The question that now arises, however, is how we measure who the ‘worst-off’ individual in society would be. Simply knowing that primary goods are universally beneficial will not always be enough for agents to rank social states. Suppose, for simplicity, that there are only 3 primary goods: rights, liberties and income. We can only determine if one individual’s bundle is better than another’s if one has more of every good, for example, x₁ = (3,5,3) and x₂ = (2,4,1), where (a,b,c) = units of each primary good. In this case, it is clear that x₁ is the better-off individual. However, problems arise when comparisons are to be made between bundles such as x_{1 =} (3,5,3) and x_{3 =} (4,4,4), where one individual possesses more of some types of primary goods and less of others.

Thus, the multiplicity of primary goods causes an unavoidable indexing problem that Rawls finds difficult to tackle. We shall call this indexing problem 1 – the main critical focus of this essay. Once we have a list of goods, we cannot use the list alone to determine which situations are best for individuals. As there are several primary goods, individuals might hold differing combinations of them. In order to advance a Rawlsian doctrine of justice that measures people’s condition by their holdings of primary goods, we need an index that assigns comparative weightings to each primary good. If this is achieved then we may, for any two individuals, determine which is more disadvantaged overall with respect to her primary goods holdings. That is, by creating an index, agents are able to evaluate a bundle of primary goods and assign a value of utility to that bundle. For example, we could now have utility functions:

f (3,5,3) = 3 and f (4,4,4) = 5

Here, the first bundle of goods has a utility of 3, whilst the second combination has a utility of 5.

The Measurability Problem and ‘Indexing Problem 2’

Before we go on to discuss the operational difficulties of assigning an index, it is important to mention two problems for Rawls’ theory that arise at this point. The first is the capability critique discussed at the start of this essay. If we have even the slightest commitment to the idea that individuals differ in their ability to convert primary goods into what is good for them, then an accurate value for utility can never be calculated at this stage. That is, the interindividual differences of persons mean that interpersonal utility comparisons cannot be made in this manner. The second is what I call the measurability problem. This is a fundamental issue that occurs prior to indexing problem 1. So far we have assumed that values may be assigned to quantities of each primary good. For example, in the example above, bundle x₁ contains 3 ‘units’ of rights, 5 of liberties, and 3 of income. It is easy to see the problem with this. Not only are the three types of primary goods difficult to compare with each other (how do we measure a unit of income in comparison to one of liberty?), there are also internal comparison complications. That is, some types of a particular primary good may deliver more utility than other types of the same primary good. This is more of a problem for qualitative primary goods such rights and liberties than it is for income. For example, people may generally value the right to education more than the right to free speech. Indeed, it is difficult to envisage an operational method by which we can convert a variety of rights into ‘units’ of rights.

Assuming, for now, that indexing problem 1 can be solved with the creation of an index, I will now briefly discuss a second issue that arises for Rawls at this point – indexing problem 2. To recap, the need to create an index is to allow for agents to: a) aggregate the utility of each, individual bundle of primary goods, and b) proceed to rank social states. For step b), Rawls asserts that agents, ignorant of facts about themselves or society, will opt to implement the difference principle with its maximin strategy. Much of the criticism levelled at Rawls’ principles has come at this point in his work, with scholars such as Harsanyi (1975) refuting maximin, instead proposing that agents would follow a strategy of expected-utility maximisation. Whilst Harsanyi appreciates that maximin would be one of two possible decision-rules for ranking social states in the OP, he argues that expected-utility maximisation is the more rational option. This is because it is unreasonable to assume that agents in the OP would be so risk averse that they (incorrectly) assign a near-unity probability to the possibility that they would end up as the worst-off individual in the society (p.599). Stating the two strategies formally:

For a social state X_k = (x₁,x₂,…,x_n) = (5,4,…,2), agents must decide how preferable it is compared with other possible social states. For example, X₂ > X₁ > X₃ > … (i.e. X₂ is most preferable).

Under a maximin strategy agents would simply rank social states according to the utility of the worst-off individual in each state, from highest utility to lowest.

Under a strategy of expected-utility maximisation, however, agents would use the following to calculate the best social state:

Depending on which strategy agents use for making preferences, the preference ordering of social states will be different; meaning that what is regarded the ‘best’ social state can differ. Thus we can see that this represents another problematic indexing issue for Rawls, and another which has divided social theorists.

Difficulties in Constructing an Index that Resolves Indexing Problem 1

Returning to indexing problem 1, Arneson (1990: 445) acknowledges the difficulties in conceiving a method by which we may aggregate an individual’s holdings of primary goods into a composite measure of the value of his holdings. He goes on to accuse Rawls of sidestepping the indexing problem, citing Rawls’ claim that those with the least wealth and income will also hold less of all other primary goods (Rawls, 1999: 94). Arneson proposes that there are only two possible answers to the indexing problem, neither of which would be palatable for a Rawlsian. The first would involve making “social policy judgments based on … claims to knowledge of what is good for people, and so of what value their resources shares really have, regardless of their own opinions on the matter” (p.446). The second possible solution would accept “a subjectivist welfare standard,” meaning individuals can determine their own valuations. The first solution would clearly not sit well with a Rawlsian, given Rawls’ theory is predicated on diverse conceptions of the good. The second faces difficulty in implementation. Arneson’s indexing concerns are well-placed, as Rawls’ claim that his theory is non-welfarist hangs on this issue. Roemer (1998: 171) shares Arneson’s aggregation concerns, concurring that “Rawls proposal is incomplete, because he has not provided a determination of the profile of monotone functions that is necessary to define the generalized indices of primary goods.” Where Roemer is more forgiving to Rawls is that unlike Arneson, he does not believe the Rawlsian view must necessarily adopt an equalising welfare strategy over the maximin approach of the difference principle. That is, Roemer acknowledges the possibility of a legitimate theory being conceived whereby indices of primary goods are ordinally equivalent to welfare (p.171). In Rawls’ defence, Arneson and Roemer’s insistence over the existence of an indexing problem largely comes from the fact that Rawls’ omits discussions about how such an index would be devised. Perhaps it was too much to expect of Rawls to design a complex index proposal in TJ, in addition to the rest of the work – which is a more abstract conception.

It is commonly accepted that Rawls offers an inadequate response to indexing problem 1 [see Gibbard (1978); Blair (1988); Roemer (1998)]. In the original version of TJ, Rawls (1971: 93) recognises the existence of an index problem but develops a very limited response. He appears to dismiss the complexity of the problem by claiming that it need only be solved for the least advantaged group. “We try to do this by taking up the standpoint of the representative individual from this group and asking which combination of primary social goods it would be rational for him to prefer” (p.94). What Rawls fails to recognise, however, is that his proposed simplification of the problem involves circularity. We cannot apply the conceptual problem to the least advantaged group without first applying an index to all groups in order to determine which group is least advantaged in the first place. Is this not the most important function of a primary goods index in Rawls’ overall theory?

Later, in his revised edition of TJ, Rawls offers another solution to indexing problem 1 – one which severely conflicts with some of his moral commitments to impartiality. Rawls (1999: 172) makes clear that decisions made behind the veil of ignorance are only the first step in a process of institutional development that involves a second stage of constitutional convention and a final stage of legislative application by judges. As the stages progress, parties have an increased amount of information available to them. In his later work Justice as Fairness: A Restatement, Rawls (2001: 48) explains that the difference principle would apply at the legislative stage, and admits there would be an indexing decision at this point. The problem for Rawls is that the postponement of index determination to this stage solves no issues as he does not show why postponement is not arbitrary on his theory’s own terms. By delaying the indexing decision to this stage, it seems as if Rawls is not taking his own reasoning for the creation of the original position (OP) seriously. The OP, with the veil of ignorance as its primary feature, has the purpose of ensuring decision-making agents are impartial by removing knowledge of personal information. However, the postponement of the indexing decision to the legislative stage compromises commitments to impartiality, as “the full range of general economic and social facts” becomes available to agents (Rawls, 1999: 175). For any proponent of Rawls’ doctrine of justice, this way of dealing with indexing problem 1 cannot and should not be acceptable. Rawls cannot claim he is fully committed to neutralising morally arbitrary considerations if he is willing to delay the implementation of his most controversial principle – the difference principle – to a stage that allows personal considerations to affect agents’ decisions.

Conclusion

The problems that arise for Rawls’ primary goods approach to justice are stark. For decades, scholars have been involved in complex debates over whether Rawls is justified in his belief that primary goods are the correct metric of justice. The main opposition has come from the capabilities approach pioneered by Sen, with the debate far from settled. This essay has moved away from such discussions, instead focussing on the intricate indexing and measurability problems that Rawls’ primary goods approach faces. Indexing problem 1 is the most troubling for Rawlsians, and remains unsolved. It would not be contentious to suggest that the problem is in fact unsolvable, as Rawls’ conception of justice uses a metric consisting of categories that are not naturally comparable.

---

References

Arneson, R. 1990. “Primary Goods Reconsidered.” Nous 24: 429-54

Blair, D.H. (1988). The Primary-Goods Indexation Problem in Rawls's Theory of Justice.Theory and Decision. 24 (0), pp.239-252

Gibbard, A. (1979). Disparate goods and Rawls' difference principle: A social choice theoretic treatment.Theory and Decision. 11 (3), pp.267-288

Harsanyi, J.C.. (1975). Can the Maximin Principle Serve as a Basis for Morality? A Critique of John Rawls's Theory.The American Political Science Review. 69 (2), pp.594-606

Osborne, M.J. (2004).An Introduction to Game Theory. Oxford: Oxford University Press

"The Priority of Right and Ideas of the Good," Philosophy and Public Affairs 17, 251-76

Rawls, J (1971). A Theory of Justice. United States: Harvard University Press

Rawls, J (1999).A Theory of Justice: Revised Edition. United States: Harvard University Press

Rawls, J (2001).Justice as Fairness: A Restatement. USA: Harvard University Press

Roemer, J.E (1998).Theories of Distributive Justice. USA: Harvard University Press

Sen, A. (1979). Equality of What?.The Tanner Lecture on Human Values. pp.198-220

Sen, A. (1990). Justice: Means Versus Freedoms.Philosophy & Public Affairs. 19 (2), pp.111-121